Abstract
This paper presents a development of the stochastic driver model based on analysis of the driving characteristic. If proper human cognition/error mechanisms are included in driver model’s states, it is possible to reproduce real driving behavior. In order to develop the stochastic state for the driver model, driver’s characteristics was analyzed by using the manual driving data. By using this analysis, states for the stochastic driver model were defined as combination terms of the nominal states and error states which are caused by human perceptual limitation, time delay, and distraction, etc. The stochastic driver model is composed of 3 parts: 'recognition', 'decision' and 'action'. Consequently, the stochastic driver model can generate the driver’s stochastic characteristics in longitudinal and lateral dynamics. Also, to validate the stochastic driver model, comparison between simulation results of the stochastic driver model and manual driving data was conducted. Consequently, the proposed driver model presented in this study can be used into a closed-loop simulation and development of the vehicle's active safety system.
Keywords: stochastic driver model, driver characteristic, static driver model, nominal state, error state